slide chương 5 tài chính hành vi lucy f ackert richard deaves

Perception and processing constraints• Expectations influence perceptions.• People see what they want to see.• People experience cognitive dissonance when they simultaneously hold two th

Chapter 5: Heuristics and Biases

Powerpoint Slides to accompany Behavioral

Finance: Psychology, Decision-making and Markets by Lucy F Ackert & Richard Deaves

Perception and processing constraints

• Expectations influence perceptions.• People see what they want to see.

• People experience cognitive dissonance when they simultaneously hold two thoughts which are psychologically inconsistent.

Perception and the frame

• Perception is not just seeing what’s there –but it is influenced by the frame:

– How tall is that sports announcer?

– Halo effects: Someone who likes one outstanding

attribute of an individual likes everything about the individual

– Primacy vs recency effects

• With emotion playing a role

– It is prone to self-serving distortion (hindsight bias)

• Heuristics or rules-of-thumb: decision-making shortcuts.

• Necessary because the world, being a

complicated place, must be simplified in order to allow decisions to be made.

• Heuristics often make sense but falter when used outside of their natural domain

Type 1 & 2 heuristics

• Type 1: Autonomic and non-cognitive, conserving on effort.

– Used when very quick choice called for– Or when it’s “no big deal”

• Type 2: Cognitive & requiring effort.

– Used when you have more time to ponder

• Type 2 can overrule Type 1.

Self-preservation heuristics

• Hear a noise with an unknown source?

– Move away till you know more

• Food tasting off?

– Stop eating it

• These make good sense.

• Other heuristics, which are more cognitive, are related to comfort with the familiar…

Example: Diversification heuristic

• Observe people at a buffet…

– Many people are trying a bit of everything– Nobody wants to miss out on something

• Diversification sometimes comes naturally.

Example: Ambiguity aversion

• In experiments, people are more willing to bet that a ball drawn at random is blue if they

know the bag contains 50 red and 50 blue.

– Than if they know a bag contains blue and red balls in unknown proportions

• Lesson: people are more comfortable with risk vs uncertainty (ambiguity).

Example: Status quo bias or endowment effect

• What you currently have seems better than what you do not have.

• Experimental subjects valued something that they possessed (after it was given to them) more than they would have if they had to consciously go out and buy the item.

Example: Information overload

• Experiment involving tasting jams and jellies in a supermarket.

• Treatment 1: Small selection.• Treatment 2: Large selection.• Which attracted more interest?

– Treatment 2.

• Which lead to more buying?

– Treatment 1.

• People judge probabilities “by the degree to which A is representative of B, that is, by the degree to which A resembles B.”

– A can be sample and B a population OR A can be a person and B a group OR A can be an event/effect and B a process/cause

• Behaviors associated with representativeness:

Conjunction fallacy

• Which seems more likely?

– a Jane is a lottery winner.

– b Jane is happy lottery winner.

• Many pick b, but a must have a higher probability, as

a Venn diagram clearly shows.• Problem: conjunction fallacy.

Conjunction fallacy: Venn diagram

Base rate neglect and Bayes’ rule

•pr(B|A) = pr(B) * [pr(A|B) / pr(A)]

• This is a way of updating your probability estimate based on new information.

• You have a barometer that predicts weather.• Example:

–pr(rain) = pr(R) = 40%–pr(dry) = pr(D) = 60%

–pr(rain predicted | rain) = pr(RP|R) = 90%–pr(rain predicted | dry) = pr(RP|D) = 2.5%

Bayes’ rule cont.

• Best prediction of tomorrow’s weather

without looking at barometer is prior (base rate) distribution: you would say 40% chance of rain.

• What should you predict when barometer predicts rain? That is, what is probability of rain conditional on rain being predicted?

•pr(R|RP) = pr(R) * [pr(RP|R) / pr(RP)]

Bayes’ rule cont ii.

• We first need to work out pr(RP).• This equals:

Bayes’ rule cont iii.

• Next work out pr(RP  D).

• Begin with conditional probability:pr(RP|D) = pr(RP  D) / pr(D) • Re-arrange:

pr(RP  D) = pr(RP|D) * pr(D)= 025 * 6 = 015• Therefore pr(RP) = 36 + 015 = 375

• Note that the barometer (conservatively) predicts rain less than it actually rains.

Using Bayes’ rule

• Best prediction of tomorrow’s weather without looking at barometer is prior (base rate)

distribution: you would say 40% chance of rain.• What should you predict when barometer

predicts rain?

pr(R|RP) = pr(R) * [pr(RP|R) / pr(RP)]= 4 * (.9 / 375) = 96

• Base rate underweighting would imply that you believe there is a higher than 96 chance of rain conditional on rain being predicted.

Hot hand phenomenon

• Sometimes people feel that

distribution/population should look like sample, but sometimes they feel sample should look like distribution/population.

– Former is especially true if people aren’t sure about nature of distribution/population.

– As in hot hand phenomenon in sport:

• In basketball, it is erroneously thought that you should give ball to hot player

• “We are due for heads.”

– Winning lottery numbers are avoided based on mistaken view that they are not likely to come up

Overestimating predictability

• Tendency to underestimate regression to

mean – amounts to exaggerating predictability.• GPA example: subjects were asked to predict

GPA in college from high school GPA of entrants to the college.

– High school average GPAs: 3.44 (sd = 0.36); GPA achieved at college was 3.08 (sd = 0.40)

– One student was chosen: high school GPA of 2.2 – People underestimated mean regression for this low-

Biases related to representativeness

– Most people will come up with a low

estimate: anchored on product of first 4 or 5.

– A bit better (but still too low) with:8 * 7 * 6 * 5 * 4 * 3 * 2 * 1

Anchoring bias: Example of anchoring to irrelevant info

• Wheel with numbers 1-100 was spun – Subjects were asked:

• 1 Is the number of African nations in the UN more or less than wheel number?

• 2 How many African nations are there in the UN?

– Answers were highly influenced by wheel:

• Median answer was 25 for those seeing 10 from wheel.• Median answer was 45 for those seeing 65 from wheel.

– Grasping at straws!

• Which is right?

– Perhaps both depending on situation…

Anchoring vs representativeness ii.

• It is argued that people are “coarsely calibrated.”

• Suppose morning forecast is for sun Day starts sunny You go on a picnic.

– Some dark clouds start to move in

– You are anchored to prior view and discount clouds

– More dark clouds: the same thing

Anchoring vs representativeness iii.

– Even more dark clouds.

– Now you coarsely transition – thinking that “it’s

going to rain for sure!”

– What is reality? Never 0% or 100% New

information should alter probabilities but a flop doesn’t make sense.

flip-• Coarse calibration has been used to explain tendency for prices to trend and eventually reverse.

Preview of financial errors from heuristics and biases

• Expectations influence perceptions:

– If most people are saying good/bad things about company, you will “find” good/bad things

• It has been argued that cognitive dissonance can:

– Explain why people don’t exit poorly-performing mutual funds

Preview of financial errors from heuristics and biases ii.

Preview of financial errors from heuristics and biases iii.

• Representativeness (and halo effects)

– “Good companies are good stocks” thinking may lead to value advantage

• Recency

– May explain chasing winners

• Anchoring and slow adjustment coupled with representativeness

– May explain momentum and price reversal